v1.1.15

• Do not fail the CSV import when the python sniffer raise an error. the default python sniffer has some issues. This commit add test case when the original file has only single column...

v1.1.14

• New date format (%d %b %Y)
• Fix unixTimestamp for INF float
• Improve CI

v1.1.13

• Wrong encoding exception
• Could not determine delimiter exception
• Fix casting exception typo
• UnixTimestamp type - catch OverflowError and ignore such cases

v1.1.12

• Enforce the IPv4 type to inference only string inputs
• Improve the error message for cases when string value found in integer column

v1.1.11

• Hotfix: Catch OSError while trying to convert numeric data to unixtime

v1.1.10

• Hotfix: Fix cases when all the column data is NAN. The value recognized as timestamp instead of float and raise an exception.

v1.1.9

• Hotfix: unix_timestamp get exception when the content is from string type
• Added: pipeline badge

v1.1.8

• Infer string-like numeric numbers using locale ("1,234.56")
• Optional encoding and locale parameters added to the Shmessy constructor
• Added exceptions handler
• Removed unsupported strftime formats (Leading zeros formats)
• Added timestamp unit-tests
• Fix: support INF/NAN values for float type

v1.1.7

Hotfix - Reading CSV file from a buffer - stream.readlines() reading all of the data instead just a sample...

v1.1.6

• Fix the infer for object type as input for Shmessy
• Remove casting type logic
• Add more timestamp formats
• Add debug logs while inferring base types
• Improve boolean infer